EP0339365B1 - Method and circuit arrangement for reducing flimmer in a television receiver - Google Patents

Abstract

To improve the picture quality in a television receiver which displays the received television signal in accordance with the line interlace method, frame stores are increasingly used. The remaining system-related flicker disturbances caused by the line interlace method require different signal processing for stationary and moving frame sequences in known flicker reduction processes, in which the receiver switches from flicker-free to motion-correct 100-Hz field repetition rate even with a relatively slight movement. To reduce system-related line flicker disturbances with line interlace reproduction, the signals contained in the frame store are in each case divided by vertical filtering in the television receiver into a vertical high-frequency and low-frequency signal as determined by the position frequency, these signals are differently processed in dependence on movement and the processed high-frequency and low-frequency signals are reproduced with twice the vertical frequency in line interlace. The flicker reduction method according to the invention can be used in all television receivers in which the television signal is reproduced at twice the vertical frequency in line interlace. <IMAGE>

Description

The invention relates to a method for reducing flicker in a television receiver according to the preamble of patent claim 1.

The image quality can be improved in television receivers by using image memories. The image memory can not only be used to suppress noise or cross-color interference or to expand functions (e.g. image in the image), but also to reduce system-related flicker interference. Flicker disorders include line flickering, line wandering, edge flickering and large-area flickering. To reduce the large-area flicker, the interlace method was introduced, in which a full image is split into two partial images, one partial image containing all odd lines and the other partial image containing all even lines. A number of different flicker reduction methods with 100 Hz field frequency while maintaining the interlacing method are known. 139 "or" Broadcasting communications, vol. 31 (1987), number 2, pp. 75 - 82 "are described in detail. These comparative considerations clearly show that a combination of the known flicker reduction methods is preferred for an effective elimination of line flicker disturbances, the transition between the known flicker reduction methods being carried out using a motion detector.

With the help of interpolation methods, the effort for the motion detector can be reduced, but the simultaneous reduction of motion disorders and flicker disorders requires a great deal of effort in signal processing. The changeover between flicker-free field display sequence and movement-correct field display sequence must be carried out even at low movement speeds.

The invention is based on the object of specifying a method for flicker reduction in a television receiver in such a way that the transition from line flicker-free to movement-correct field display sequence can only be carried out at relatively high movement speeds.

This object is achieved by a flicker reduction method with the features of patent claim 1.

The flicker reduction method according to the invention is based on the knowledge that line flicker effects are predominantly influenced by vertical height components in accordance with the spatial frequency. By means of the vertical filtering of the signals contained in the image memory, signal processing can now be carried out for the height and depth signals thus obtained in such a way that the height signals are displayed without line flickering and the depth signals are reproduced correctly. As a result, in a surprisingly simple manner, the transition to a movement-correct field display sequence is only necessary for rapidly moving image areas.

In the embodiment of the flicker reduction method according to claim 2, the height signals are offered approximately with the flicker quality of a progressively reproduced frame, i.e. that is, line-flicker-free image reproduction is achieved. This also largely compensates for alias errors. The depth signal shows moving objects with their essential limitations in the correct direction of movement. In order to avoid fluttering edges, there is a soft reversal between the field display sequences for still and slowly moving image areas or rapidly moving image areas.

The flicker reduction method according to claim 3 requires little signal processing, since the vertical re-interpolation is only to be carried out for the depth signals.

The weighted and temporal interpolation of the depth signals according to claim 4 results in a good representation of movement at lower and medium speeds. The smearing of the edges of moving objects occurs only in the horizontal direction due to the height-depth separation, i.e. with vertical edges and in a weakened form with diagonal edges. The motion blur that occurs can, however, be spatially corrected in a simple manner by a flicker reduction method according to patent claim 5 or 6.

The circuit arrangement for carrying out the method according to claim 7 requires little circuit complexity, in particular when the motion detector is implemented, and allows the complementary filter edges to be determined appropriately.

Preferred configurations of the circuit arrangement are specified in further patent claims.

• Fig 1 das Blockschaltbild einer ersten Ausführungsform zur Durchführung des erfindungsgemäßen Verfahrens,
• Fig 2 das Blockschaltbild einer Ausführungsform für einen Frequenzbandselektor,
• Fig 3 schematisch die Signalverarbeitung für eine 100-Hz-Halbbildwiedergabefolge mit Vollbildwiederholung der vertikalen Höhensignale und Halbbildwiederholung der vertikalen Tiefensignale,
• Fig 4 das ortsfrequente Spektrum der komplementären, vertikalen Höhen-Tiefensignal-Trennung,
• Fig 5 das Blockschaltbild einer Ausführungsform für einen bewegungsadaptiven Frequenzbandselektor,
• Fig 6 schematisch die Signalverarbeitung für eine 100-Hz-Halbbildwiedergabefolge mit Vollbildwiederholung der vertikalen Höhensignale und zeitlicher Mittelung der vertikalen Tiefensignale,
• Fig 7 das Blockschaltbild einer Ausführungsform für einen gesteuerten Tiefeninterpolator mit Kantenversteilerung,
• Fig 8 den Leuchtdichteverlauf bei gewichteter linearer Interpolation,
• Fig 9 das Zeitdiagramm für Abtastung und Gewichtung sowie die Übertragungsfunktion zur Erläuterung der Filterwirkung bei linearer Interpolation und
• Fig 10 ein zweidimensionales Frequenzspektrum des Leuchtdichtesignals sowie die Übertragungsfunktion und Interpolationsübertragungsfunktion zur Erläuterung der Interpolation bei translatorischer Bewegung.

The invention is described and explained in more detail below with reference to embodiments shown in the drawing for the flicker reduction method according to the invention. It shows:

• 1 shows the block diagram of a first embodiment for carrying out the method according to the invention,
• 2 shows the block diagram of an embodiment for a frequency band selector,
• 3 schematically shows the signal processing for a 100 Hz field reproduction sequence with frame repetition of the vertical height signals and field repetition of the vertical depth signals,
• 4 shows the spatial frequency spectrum of the complementary, vertical height-depth signal separation,
• 5 shows the block diagram of an embodiment for a motion-adaptive frequency band selector,
• 6 schematically shows the signal processing for a 100 Hz field display sequence with frame repetition of the vertical height signals and temporal averaging of the vertical depth signals,
• 7 shows the block diagram of an embodiment for a controlled depth interpolator with edge distribution,
• 8 shows the luminance profile with weighted linear interpolation,
• 9 shows the time diagram for sampling and weighting as well as the transfer function for explaining the filter effect with linear interpolation and
• 10 shows a two-dimensional frequency spectrum of the luminance signal and the transfer function and interpolation transfer function to explain the interpolation during translatory movement.

1 shows the block diagram of a first embodiment for implementing the method according to the invention. A demultiplexer D is connected to an image memory SP, in which the incoming signal Si is stored field by field (fields A, B). A complementary, vertically interpolating filter F, an interpolator 1 and a motion detector BD are connected to the image memory SP. Using the filter F, the fields A, B are divided into height-depth signals, for example A _oH , A _OT , and fed to a motion- _adaptive control ST via a switch S. The switching frequency of the switch S corresponds to twice the vertical frequency T. The motion-adaptive control ST is also connected to the motion detector BD, a monitor M and an image repeat switch W, to the input of which the interpolator I is connected.

In the case of quiet and moving parts of the image up to approximately 2 to 3 pixels, the upper branch with the filter F and the switch S is selected by the motion-adaptive control ST with the aid of the motion detector BD using the motion detector BD. In the case of even higher movement speeds, the movement-adaptive control ST controls the lower branch with the interpolator and the image refresh switch Wang. The height signals A _OH, B _OH are on the upper branch of the two fields A, B in a flicker-free 100 Hz field reproduction sequence A _OH, B _OH, A _OH, B _OH and the depth signals A _OT, B _OT or interpolated depth signals A _OTi , B _{OTi of} the two fields A, B in a 100 Hz field _{reproduction sequence} A _OT , A _OTi , B _OTi , B _OT fed to the monitor M. The letter i indicates a temporal interpolation between the field grids, which can be carried out using suitable filters. The fields A, B or fields A _i , B _i resulting from interpolation are displayed on the monitor in a 100 Hz field display sequence A _o , A _Oi , B _Oi , B via the lower branch with interpolator I and image repeat switch W in the case of rapidly moving image areas _{Reproduced O} I is also intended to indicate a temporal interpolation between the field grids, which can be carried out using suitable filters.

Fig. 2 shows in detail the structure of the complementary, vertically interpolating filter F and the switch S (frequency band selector FBS). The filter F has a vertical height signal filter F1 and a vertically interpolating depth signal filter F2. A first switch S1 is connected to the height signal filter F1 and a second switch S2 is connected to the depth signal filter F2. The inputs of the height signal filter F1 and the depth signal filter F2 are connected to the image memory SP. At the output of the first and second switches S1, S2, the respective field display sequence (100 Hz clock T1 or T2) for stationary and slowly moving image areas can be tapped, which is fed to the monitor M via an adder circuit AD. The field reproduction sequences occurring at the output of the first and second switches S1, S2 are described and explained in more detail below with reference to FIG. 3.

3 schematically shows the signals occurring in the flicker reduction method in the different signal processing stages. The incoming interlaced signal shown in the first line of FIG. 3 (fields A _o , B _O , A ₁ , B ₁ etc.) is divided by a complementary filtering into a vertical high and low component according to the spatial frequency (cf. second line 3, height signal A _OH , B _OH , A _1H , B _1H etc. and depth signal A _OT , B _OT , A _1T , B _1T etc.).

For stationary and slower moving image areas, the height signals A _OH , B _{oH of} the two fields are _reproduced in a flicker-free 100 Hz field _{reproduction sequence} A _OH , B _OH , A _OH , B _OH . This also ensures line flicker-free image reproduction, since the vertical height signals A _OH , B _oH largely responsible for line flicker are _{offered to} the eye - in the heights - approximately in the flicker quality of a progressively reproduced full image. Alias errors of the fields A _o , B _O , etc., which follow one another at 10 ms intervals, are largely compensated for.

The depth signals A _OT , B _OT or interpolated depth signals A _OTI , B _{OTi of} the two fields A _O , B _O are reproduced in a movement- _correct 100 Hz field reproduction _sequence A _OT , A _OTI , B _OTi , B _OT . The temporal Interpolation can be carried out in a simple manner, for example by line averaging, since this is only to be carried out for the depth signal A _OT , B _OT etc. This embodiment of the vertical repetition of depth is particularly favorable for a movement display, since there is no jump back in time. In particular, the vertical depth signal A _or , B _OT etc. (which also contains all horizontal heights) shows moving objects with their essential limits in the correct direction of movement. The fine details, signaled in the vertical heights A _OH , B _oH etc., which can cause line flicker, are reproduced in the flicker-free 100 Hz field _{display sequence} A _OH , B _OH , A _OH , B _OH .

For the assessment of the flicker reduction method, the spectral consideration of the vertical direction and the time is sufficient, since only in these two directions is scanned discretely, while a continuous reproduction takes place in the horizontal direction. Due to the temporally sequential scanning of the lines, the time frequency f ^I and the spatial frequency f ^{y are} not independent of one another, approximately the spectra of the vertical direction y and the time t can be combined multiplicatively into two-dimensional spectra.

4 shows the frequency response for an embodiment of a vertical, complementary filter. The spatial-frequency spectra of the two fields A _o and B _o are periodically in the direction of the spatial frequency axis f ^y because of the vertical scanning. The alias errors contained in the two fields A _o and B _o are compensated for by the opposite sign.

The high-pass filtering by means of the vertical height signal filter F1 with the frequency response H _H produces signals which predominantly contain the alias errors. As a result, alias compensation and flicker reduction can be carried out for the height signals A _OH , B _OH . The low-pass filtering, which is complementary to the high-pass filtering, by means of the vertical depth signal filter F2 and the frequency response H _T, supplies predominantly the same-sign basic signals in the fields A _o and B _o , so that the movement can be represented with great fidelity.

With larger movements, the image contents recorded at different times are no longer the same, so that the field spectra differ in shape and sign. Even with ideal integration, alias errors would no longer be compensated for. Artifacts would be visible on fine horizontal details, ie at high vertical spatial frequencies fy. In order to avoid this, the motion-adaptive control ST switches from the upper to the lower branch and the fields A _O , B _O or interpolated fields A _Oi , B _Oi in a movement-correct 100 Hz field reproduction sequence A _O , A _Oi , B _Oi , B _O reproduced. The soft reversal prevents fluttering edges from occurring.

In order to reduce the system-related line flicker interference, an optimization of the complementary filter edges of vertical height signal filter F1 and vertically interpolating depth signal filter F2 is necessary. This optimization, with regard to the signals to be processed, can be simplified if motion-adaptive control of the filter edges is carried out. An embodiment for a motion adaptive frequency band selector is shown in FIG. The height signal filter F1 and the depth signal filter F2 are connected to a motion detector with filter control BDF which, depending on the speed of movement, shifts the pass bands of the two filters F1, F2 in such a way that the field or sequence for the height or depth signal is reversed.

In the case of stationary and very weakly moving image parts, the signal processing of the two fields is carried out in accordance with the signal processing for the height signal, i.e. the edge of the vertical height signal filter F1 is zero at the cutoff frequency. As the speed of movement increases, this flank is then shifted continuously or in steps to higher frequencies, until finally at high speeds the signal processing for the two fields is carried out in accordance with the movement-correct signal processing for the depth signal. In this way, motion-adaptive control is made possible, which optimizes motion reproduction and flicker reduction at every motion speed.

6 schematically shows the signal processing of a 100 Hz field display sequence, in which the vertical height signals, which are vertical according to the spatial frequency, are reproduced without line flicker and for the depth signals, signal processing with weighting and temporal interpolation is carried out. In signal processing, vertical height-depth signal separation, temporal, linear interpolation of the vertical depth signals and motion-adaptive addition of the vertical height signals are carried out.

The incoming signal is first divided into a vertical high and low signal by a complementary filtering (cf. first and second line of FIG. 6). The vertical height signal is reproduced for stationary and weakly moving parts of the picture in the flicker-free 100 Hz field display sequence A _OH , B _OH , A _OH , B _OH . As a result, line flicker-free image reproduction is achieved, since the vertical height signals largely responsible for line flicker are offered to the eye approximately in the flicker quality of a progressively written full image. The alias errors of the fields that follow one another at 10 ms intervals largely compensate each other. For the weighting and temporal interpolation of the tie Different signals are conceivable, with a view to a good movement representation, signal processing is preferably carried out according to the following equation:

The resulting signals are shown in the signal processing diagram of Fig. 6 (third line). The temporal interpolation for the vertical depth signals (i.e. with the horizontal depth and height signals) ensures a good representation of movement at lower and medium speeds. As with all temporally interpolating methods, the edges of moving objects are smeared without further measures, but due to the height-depth signal separation, this only in the horizontal direction, i.e. for vertical and - weaker - for diagonal edges. By means of motion-adaptive signal processing with edge distribution, the signal processing according to equation (1) can, however, be carried out at speeds of approximately 1 to 2 pixels per field.

FIG. 7 shows an embodiment for motion-adaptive signal processing with edge distribution. The switch S2 is replaced by a depth interpolator TI and a motion-adaptive edge distribution circuit KV connected to its output. The depth interpolator TI is connected to the motion detector with filter control BDF. The functioning of the spatial correction with temporal movement smearing is described and explained in more detail below with reference to FIGS. 8, 9 and 10.

8 shows a vertically moving vertical edge, the luminance profile g ₁ (x, t ₁ ) or g ₂ (x, t ₂ ) being indicated at the time t ₁ or t ₂ . As FIG. 8 clearly shows, edge blending occurs as a result of the weighted linear interpolation. The filter effect of such a linear interpolation can be described by convolution with a triangular impulse response shown in FIG. 9, a filter effect occurring in the frequency range, which can be described by the frequency response H (f) = T _{o Si} ² (πfT _O ).

The following then applies to the edge moving translationally in the x direction:

Using the Fourier transform law, we get for the moving edge:

geschert. Eine lineare zeitliche Interpolation erzeugt eine Gewichtung mit der Interpolationsübertragungsfunktion H_t (f^t) für das gescherte Spektrum und wirkt räumlich mit der Übertragungsfunktion H_d (f^d) in Richtung des gescherten Spektrums.In order to simplify the observation, the line scan and the line jump are now neglected and only the time scan is considered. The periodic representation given in FIG. 10 results for the spectrum in the f ^X , f ^t plane of the signal according to equation (3). The translationally moving spectra are along the straight line

sheared. A linear temporal interpolation generates a weighting with the interpolation transfer function H _t (f ^t ) for the sheared spectrum and acts spatially with the transfer function H _d (f ^d ) in the direction of the sheared spectrum.

If the translatory movement is such that the eye can follow the object, the movement is, as it were, compensated for the viewer. He then perceives the loss of sharpness especially at the edges with the tolerance of the eye for stationary objects. Within certain limits, the loss of sharpness can be compensated for with a purely spatial, linear or non-linear increase in the f ^x direction, inverse to the transfer function H _d (fd).

The edge elevation is adaptive in the direction of movement and proportional to the speed of the speed vector. By restricting the movement-dependent steepening to vertical depth signals and adding the vertical height signals later, motion vector estimation is possible with little effort, in particular a one-dimensional, horizontal frequency response increase or nonlinear edge distribution in the same direction leads to satisfactory results. Further embodiments for this are given by a weighting of the frequency increase approximately proportional to the speed amount or a weighted edge detail signal addition approximately proportional to the speed amount.

Claims (10)

1. Flicker-reduction method for a television receiver, in which the received television signal is temporarily stored in fields in a frame store (SP) of the television receiver, the fields being subjected to a signal processing and being reproduced at twice the field frequency by the interlaced scanning method, characterized in that, as a result of a vertical filtering of the signals contained in the frame store (SP), they are divided up into a vertical high-frequencies signal and low-frequencies signal (A_OH, B_oH and A_OT, B_OT) as a function of the spatial frequency and in that the high-frequencies and low-frequencies signals (A_OH, B_oH and A_OT, B_OT) thus obtained are subjected to a signal processing with different reproduction sequences of high-frequencies and low-frequencies signals of the fields.

2. Flicker-reduction method according to Claim 1, characterized in that, for stationary and slowly moving picture areas, the high-frequencies signals (A_OH, B_OH) of the two fields, on the one hand, are reproduced in a flicker-free 100 Hz field reproduction sequence (A_OH, B_OH, A_OH, B_OH) and, on the other hand, the low-frequencies signals (A_OT, B_OT) with low-frequencies signals (A_OTi, B_OTi) of the two fields interpolated therefrom are reproduced in a motion-adapted 100 Hz field reproduction sequence (A_OT, A_OTi, B_OTi, B_OT) under the control of a motion detector (BD), and in that, in the case of rapidly moving picture areas, the two fields (A_o, B_o) and the fields (A_o;, B_Oi) interpolated therefrom are reproduced in a motion-adapted 100 Hz field reproduction sequence (A_o, A_Oi, B_Oi, Bo).

3. Flicker-reduction method according to Claim 1, characterized in that a vertical interpolation is carried out by line averaging of the low-frequencies signals (A_OT, B_OT).

4. Flicker-reduction method according to Claim 1, characterized in that the low-frequencies signals (A_OT, B_OT) are weighted and time-interpolated during the signal processing.

5. Flicker-reduction method according to Claim 4, characterized in that a nonlinear edge slope increase of the low-frequencies signals (A_OT, B_OT) is carried out adaptively in the direction of the motion vector and proportionally to its velocity modulus.

6. Flicker-reduction method according to Claim 5, characterized in that, to reduce the motion-unsharpness, a nonlinear edge slope increase of the low-frequencies signals (A_OT, B_OT) is carried out in the direction of the horizontal positional coordinate.

7. Circuit arrangement for carrying out the method according to one of the preceding claims, characterized in that the incoming signal is temporarily stored in fields by means of a demultiplexer (D) fitted in the television receiver and a frame store (SP) connected thereto, in that, for the purpose of division into high-frequencies signal and low-frequencies signal (A_oH, B_oH and A_OT, B_OT), a vertically interpolating filter (F) is connected to the frame store (SP), in that the filter (F) is connected to a changeover switch (S) which feeds the high-frequencies signal and the low-frequencies signal (A_OH, B_oH and A_OT, B_OT, A_OTi, B_OTi) to a motion-adaptive controller (ST) in different field reproduction sequences, and in that the motion-adaptive controller (ST) is connected, on the one hand, to a motion detector (BD) connected to the frame store (SP) and is connected, on the other hand, to the frame store (SP) via a frame repetition switch (W) and an interpolator (I).

8. Circuit arrangement according to Claim 7 in combination with Claim 2, characterized in that the vertically interpolating filter (F) comprises a vertical high-frequencies signal filter (F1) and a vertically interpolating low-frequencies signal filter (F2) and in that a first switch (S1) is connected to the high-frequencies signal filter (F1) and a second switch (S2) is connected to the low-frequencies signal filter (F2), the respective field reproduction sequences for stationary and slowly moving picture areas being capable of being tapped off at the output of the first and second switch (S1 and S2).

9. Circuit arrangement according to Claim 8, characterized in that the high-frequencies signal filter (F1) and the low-frequencies signal filter (F2) are connected to a motion detector with filter control (BDF), which detector carries out a displacement of the pass ranges of the two filters (F1, F2) as a function of the movement velocity in such a way that switching is carried out between the field reproduction sequences comprising high-frequencies or low-frequencies signals.

10. Circuit arrangement according to Claim 8, characterized in that a low-frequencies interpolator (TI) is connected to the low-frequencies signal filter (F2) and a motion-adaptive edge slope increase circuit (KV) is connected to its output.

Images